On 02/27/2016 10:37 AM, clipka wrote:
> Am 27.02.2016 um 13:54 schrieb ThH:
>
>> 3 different pics by changing 1 line of the original optics.pov...
>
> What really intrigues me, and what I would like to ask you to try and
> investigate, is why only /some/ of the light rays are affected; why
> don't /all/ the blue and /all/ the red rays end up blazingly white?
>
> There must be /some/ reason for this, and I reckon that it might lead to
> a useful clue.
>
Christoph,

I think it has something to do with the light type.

// <--- Fails 3.7.1-alpha.8454683 Ubuntu 14.04  (OK with 3.7.0-stable)
// light_source {<-150, 0.5, 0>, color rgb < 1.2, 1, 1.5>
//     spotlight radius 0.3 falloff 0.35 point_at < 0, 0.5, 0>
//     photons {refraction on reflection on}
// }
// <--- OK Ubuntu 14.04 3.7.0 & 3.7.1
    light_source {<-150, 0.5, 0>, color rgb < 1.2, 1, 1.5>
        photons {refraction on reflection on}
    }

Perhaps someone else can verify ?

I extended the blocking wall for the point light, but not strictly needed.

Bill P.

Post a reply to this message

Am 27.02.2016 um 17:17 schrieb William F Pokorny:

> Christoph,
>
> I think it has something to do with the light type.
>
> // <--- Fails 3.7.1-alpha.8454683 Ubuntu 14.04  (OK with 3.7.0-stable)
> // light_source {<-150, 0.5, 0>, color rgb < 1.2, 1, 1.5>
> //     spotlight radius 0.3 falloff 0.35 point_at < 0, 0.5, 0>
> //     photons {refraction on reflection on}
> // }
> // <--- OK Ubuntu 14.04 3.7.0 & 3.7.1
>     light_source {<-150, 0.5, 0>, color rgb < 1.2, 1, 1.5>
>         photons {refraction on reflection on}
>     }
>
> Perhaps someone else can verify ?

Tried it...

Yes. The resulting image differs, bit still not right.

I've attached my rendering.

Post a reply to this message

Attachments:
Download 'optics_sandbox.png' (50 KB)

Preview of image 'optics_sandbox.png'

Am 27.02.2016 um 16:37 schrieb clipka:
> Am 27.02.2016 um 13:54 schrieb ThH:
>
>> 3 different pics by changing 1 line of the original optics.pov...
>
> What really intrigues me, and what I would like to ask you to try and
> investigate, is why only /some/ of the light rays are affected; why
> don't /all/ the blue and /all/ the red rays end up blazingly white?
>
> There must be /some/ reason for this, and I reckon that it might lead to
> a useful clue.

A new series...

How about these images?

Post a reply to this message

Attachments:
Download 'optics_sandbox_f_00.png' (26 KB) Download 'optics_sandbox_f_01.png' (26 KB) Download 'optics_sandbox_f_02.png' (35 KB) Download 'optics_sandbox_f_02_03.png' (42 KB) Download 'optics_sandbox_f_03.png' (34 KB) Download 'optics_sandbox_f_04.png' (27 KB) Download 'optics_sandbox_f_05.png' (26 KB) Download 'optics_sandbox_f_06.png' (26 KB)

Preview of image 'optics_sandbox_f_00.png'

Preview of image 'optics_sandbox_f_01.png'

Preview of image 'optics_sandbox_f_02.png'

Preview of image 'optics_sandbox_f_02_03.png'

Preview of image 'optics_sandbox_f_03.png'

Preview of image 'optics_sandbox_f_04.png'

Preview of image 'optics_sandbox_f_05.png'

Preview of image 'optics_sandbox_f_06.png'

Am 27.02.2016 um 18:08 schrieb ThH:
> Am 27.02.2016 um 17:17 schrieb William F Pokorny:
> 
>> Christoph,
>>
>> I think it has something to do with the light type.
>>
>> // <--- Fails 3.7.1-alpha.8454683 Ubuntu 14.04  (OK with 3.7.0-stable)
>> // light_source {<-150, 0.5, 0>, color rgb < 1.2, 1, 1.5>
>> //     spotlight radius 0.3 falloff 0.35 point_at < 0, 0.5, 0>
>> //     photons {refraction on reflection on}
>> // }
>> // <--- OK Ubuntu 14.04 3.7.0 & 3.7.1
>>     light_source {<-150, 0.5, 0>, color rgb < 1.2, 1, 1.5>
>>         photons {refraction on reflection on}
>>     }
>>
>> Perhaps someone else can verify ?
> 
> Tried it...
> 
> Yes. The resulting image differs, bit still not right.
> 
> I've attached my rendering.

Looks ok to me (except that the "original" light should cover the entire
height of the image's left side, but I'm pretty sure that's an entirely
different problem). Where the light streams past the block at the upper
edge of the image it's totally normal that we get bright white light.

I think it might well be worth toying around with the spotlight some more.

Post a reply to this message

Am 27.02.2016 um 18:27 schrieb clipka:

> I think it might well be worth toying around with the spotlight some more.

Did so. Images attached.

Anything looking promising?

Post a reply to this message

Attachments:
Download 'optics_sandbox_fo_1.png' (43 KB) Download 'optics_sandbox_fo_2.png' (51 KB) Download 'optics_sandbox_fo_3.png' (27 KB)

Preview of image 'optics_sandbox_fo_1.png'

Preview of image 'optics_sandbox_fo_2.png'

Preview of image 'optics_sandbox_fo_3.png'

Am 27.02.2016 um 19:07 schrieb ThH:
> Am 27.02.2016 um 18:27 schrieb clipka:
> 
>> I think it might well be worth toying around with the spotlight some
>> more.
> 
> Did so. Images attached.
> 
> Anything looking promising?

VERY promising. Whatever you did there, it has some definitive effect on
the symptoms.

Details please.

Post a reply to this message

Am 27.02.2016 um 19:10 schrieb clipka:

> VERY promising. Whatever you did there, it has some definitive effect on
> the symptoms.
>
> Details please.

Falloff:

     //spotlight radius 0.3 falloff 0.35 point_at < 0, 0.5, 0> // 1
     //spotlight radius 0.3 falloff 0.35*10 point_at < 0, 0.5, 0> // 2
     spotlight radius 0.3 falloff 0.35/10 point_at < 0, 0.5, 0> // 3

(Holds breath)

Post a reply to this message

Am 27.02.2016 um 19:22 schrieb ThH:
> Am 27.02.2016 um 19:10 schrieb clipka:
> 
>> VERY promising. Whatever you did there, it has some definitive effect on
>> the symptoms.
>>
>> Details please.
> 
> Falloff:
> 
>     //spotlight radius 0.3 falloff 0.35 point_at < 0, 0.5, 0> // 1
>     //spotlight radius 0.3 falloff 0.35*10 point_at < 0, 0.5, 0> // 2
>     spotlight radius 0.3 falloff 0.35/10 point_at < 0, 0.5, 0> // 3
> 
> (Holds breath)

While this does show a clear correlation between the falloff parameter
and the area where the error occurs, the geometry of that area is still
a puzzle to me: For instance, why is the area farther out on the red leg
than on the blue leg? And why is the area much farther out than the falloff?

Can you do me a favor and toy around with the colours, the geometry of
how the optical elements are arranged, and how this relates to the falloff?

(BTW, when you post results, please include the details right away. Me
having to ask for them unnecessarily bogs down the whole process.)

Post a reply to this message

Ha!

Does this look familiar to anyone?

It turns out that the root problem _is_ also present in the Windows
version -- but it's masked by a peculiarity of the run-time library's
handling of special floating-point values.

What happens is that _somewhere_ in the photon computations one of the
mathematical operations results in "not a sensible value", which carries
over into the computed colour values. How such values behave in the
conversion to the RGBE format depends on the run-time library
implementation: On Windows it results in zero values, thus "healing" the
problem. On Linux however, it results in rather high values instead.

Post a reply to this message

Attachments:
Download 'optics.png' (145 KB)

Preview of image 'optics.png'

Am 28.02.2016 um 06:02 schrieb clipka:

> While this does show a clear correlation between the falloff parameter
> and the area where the error occurs, the geometry of that area is still
> a puzzle to me: For instance, why is the area farther out on the red leg
> than on the blue leg? And why is the area much farther out than the falloff?
>
> Can you do me a favor and toy around with the colours, the geometry of
> how the optical elements are arranged, and how this relates to the falloff?

Me likes toying around with this great piece of software. I will :)
>
> (BTW, when you post results, please include the details right away. Me
> having to ask for them unnecessarily bogs down the whole process.)

Ok. Understood.

Post a reply to this message